Therapy-related acute myeloid leukemia (t-AML) is a rare but devastating complication of prior chemotherapy. That only a subset of patients treated for cancer develops t-AML suggests that these individuals may be genetically predisposed to t-AML. Alkylator-induced t-AML accounts for approximately 75% of all cases of t-AML, and is characterized by antecedent myelodysplasia and the characteristic recurrent loss of the long arm or chromosome 5 and/or chromosome 7. The long-term objective of this project is to identify genetic susceptibilities to t-AML that can be clinically translated into biomarkers for risk, or that can be used to guide therapy at the time of initial cancer diagnosis to minimize the subsequent risk of alkylator-induced t-AML. In particular, we hypothesize that genetic variation that alters the function of core stress response pathways may thereby alter cancer susceptibility. Indeed, several studies have identified genetic variants that are associated with t-AML;these studies, however, were uniformly small, reliant upon pre-existing knowledge, and limited to testing only a few plausible candidate genes. In a pilot study, we demonstrated that a genome-wide association study (GWAS) is a feasible strategy to identify genetic risk factors for t-AML, and we identified both single nucleotide polymorphisms (SNPs) and copy number variants associated with t-AML susceptibility. Here, we propose to build upon these results by undertaking a very high density GWAS to identify the complement of inherited genetic variation associated with t-AML using the Affymetrix Genome-Wide Human SNP Array 6.0. For these studies, we have amassed an unprecedented cohort of about 300 well-characterized samples from patients with t-AML, including the University of Chicago series of t-AML patients maintained in Core A of this program project. The objectives of this project will be pursued through four specific aims. In Aim 1, we will undertake a GWAS to identify polymorphic and copy number variation associated with t-AML as compared to both cancer-free healthy control and individuals matched to cases for primary cancer, but who do not have t- AML. In Aim 2, we will undertake a novel bioinformatic analysis of these data to identify the most likely genes and pathways implicated in the pathogenesis of t-AML. In Aim 3, we will undertake a series of functional studies to clarify the role of the associated polymorphisms identified in Aims 1 and 2 on the etiology of t-AML. In Aim 4, we will determine whether genetic variants found to be associated with t-AML are also associated with de novo AML. Project 2 of this application is the genome-wide analysis of copy number changes (somatically acquired mutations in t-AML). Together, these two projects provide an unmatched opportunity to study the genetics of t-AML. Because t-AML is a unique model for the geneenvironment interactions that may drive virtually all cancer, these studies may shed significant light on the genetic contribution not only towards t-AML, but to a variety of common human cancers as well.